Re: cond_resched() and RCU CPU stall warnings

[Paul E. McKenney]

On Sun, Mar 16, 2014 at 07:14:15AM +0100, Mike Galbraith wrote:
> On Sun, 2014-03-16 at 07:09 +0100, Mike Galbraith wrote: 
> > On Sat, 2014-03-15 at 18:59 -0700, Paul E. McKenney wrote: 
> 
> > > diff --git a/kernel/sched/core.c b/kernel/sched/core.c
> > > index b46131ef6aab..994d2b0fd0b2 100644
> > > --- a/kernel/sched/core.c
> > > +++ b/kernel/sched/core.c
> > > @@ -4075,6 +4075,9 @@ int __sched _cond_resched(void)
> > >  		__cond_resched();
> > >  		return 1;
> > >  	}
> > > +	preempt_disable();
> > > +	rcu_note_context_switch(smp_processor_id());
> > > +	preempt_enable();
> > >  	return 0;
> > >  }
> > >  EXPORT_SYMBOL(_cond_resched);
> > 
> > Hm.  Since you only care about the case where your task is solo, how
> > about do racy checks, 100% accuracy isn't required is it?  Seems you
> > wouldn't want to unconditionally do that in tight loops.
> 
> (or do that in scheduler_tick()?)

There are checks in scheduler_tick(), but they have to assume that
if they interrupt kernel execution, they might have also interrupted
an RCU read-side critical section.  And in fact, the point of having
cond_resched() (sometimes) do a quiescent state is to communicate with
the existing checks invoked from scheduler_tick().

							Thanx, Paul

--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/